Path: csiph.com!weretis.net!feeder9.news.weretis.net!panix!.POSTED.spitfire.i.gajendra.net!not-for-mail
From: cross@spitfire.i.gajendra.net (Dan Cross)
Newsgroups: alt.folklore.computers
Subject: Re: Don Norman: The Truth About Unix
Date: Sun, 1 Feb 2026 15:06:54 -0000 (UTC)
Organization: PANIX Public Access Internet and UNIX, NYC
Message-ID: <10lnq6e$643$1@reader2.panix.com>
References: <10lb6d7$3she5$1@dont-email.me> <10ll48q$2vbi7$1@dont-email.me> <10llqrd$37blf$4@dont-email.me> <10ln01t$3hqko$1@dont-email.me>
Injection-Date: Sun, 1 Feb 2026 15:06:54 -0000 (UTC)
Injection-Info: reader2.panix.com; posting-host="spitfire.i.gajendra.net:166.84.136.80"; logging-data="6275"; mail-complaints-to="abuse@panix.com"
X-Newsreader: trn 4.0-test77 (Sep 1, 2010)
Originator: cross@spitfire.i.gajendra.net (Dan Cross)
Xref: csiph.com alt.folklore.computers:233865

In article <10ln01t$3hqko$1@dont-email.me>,
Peter Flass  <Peter@Iron-Spring.com> wrote:
>On 1/31/26 14:05, Lawrence D’Oliveiro wrote:
>
>> Worth also pointing out that, at the time of MS-DOS 2.0 with its
>> introduction of some bizarro-Unix features, the commonly-available PC
>> hardware was already more powerful than the original PDP-11 systems
>> that Unix was created on. So why couldn’t Microsoft (or IBM) provide a
>> full Unix-equivalent OS for PC hardware back then?
>
>No memory protection or memory mapping until the 386. There were 
>unix-like OSs for 8086, but they were terrible fragile without memory 
>protection. I don't know how it was managed on the PDP-11.

[Meta note: I wonder why people continue to engage with Lawrence
as he is an obvious troll.  Please don't feed him]

Most PDP-11 models that ran Unix had an MMU.  It wasn't a very
capable MMU, mind you, but it at least allowed the system to
protect processes from one another and protect the kernel from
processes.

John Levine has posted before about a port of Unix to the 8086,
that made use of the segmentation system to more or less protect
the OS from errant user programs; the C compiler simply did not
emit instructions to change the segmentation registers, so it
worked pretty well as I understand it.

The 80286 featured a bizarre, overly wrought "task" model in
hardware that saw little use in practice; I believe that OS/2
might have used it.  Vestiges of it persist into the modern day,
where the "TSS" (Task State Segment) has been overloaded in
64-bit "long" mode to hold a table of stack pointers that can be
used with various traps so that, say, the double-fault, NMI or
debug/breakpoint handlers can run on dedicated stacks.  There's
also some business about allowing non-privileged access to IO
ports for programmed IO from user-mode code.  Anyway.

The 80386 was designed as a 32-bit CPU for the Unix workstation
market, and supported paging natively.  I must say, all things
considered they did a pretty good job overall with the design of
the MMU and page table format.  To maintain backwards
compabibility they extended the segmentation mechanism; the
intent was that you would define segments covering the entire
32-bit address space and then more or less ignore them.

	- Dan C.